This AI Goldmine Won't Be Around For Long (Beginner Friendly)

Executive summary

Andrew Dunn argues that agencies and consultancies can use AI to deliver services at sharply lower cost while continuing to sell outcomes—not software tools or hours. He calls this model “services as software”: keep the customer relationship and promised result, but rebuild delivery around AI.

Dunn says the opportunity is temporary because service providers’ costs have fallen faster than buyers’ expectations and pricing have adjusted.

Business model and strategy

• Sell outcomes, not tools or process. For example, offer a marketing client a target such as 30 leads in 90 days, rather than access to AI or a list of marketing tasks.
• Rebuild delivery with AI. Use AI for production and follow-up while retaining human involvement for onboarding, judgment, quality control, and customer relationships.
• Choose a pricing strategy based on the deliverable:
  • Lower the price to reach customers who were previously priced out, potentially expand internationally, improve conversion, and pursue longer retention. This requires a strong, high-volume customer acquisition engine.
  • Keep the price high and capture the cost savings as margin. Dunn says this is more defensible when customers are paying for trust and expertise and cannot easily assess how the work is done.
• Be cautious with visible, easy-to-copy deliverables. Content, ads, and social media work can be compared directly, making price competition more likely. Less transparent, expertise-heavy work—such as legal, advisory, or consulting services—may support trust-based pricing.
• Rebuild delivery before changing prices. Dunn recommends first making the service more efficient with AI, then deciding whether to lower prices or retain them.

Frameworks and playbooks

• “Services as software” framework: Customers buy a result, AI performs much of the delivery, and the provider captures the gap between the old cost to serve and the new AI-enabled cost.
• Four-step implementation playbook:
  1. List the services and deliverables you currently sell.
  2. Assess how much AI can contribute to each today.
  3. Select the most valuable candidates and compress their delivery with AI.
  4. Rebuild delivery first, then choose a pricing strategy: lower-price scale or trust-based pricing.
• Four forces Dunn says support the opportunity:
  • Large existing demand for services.
  • Lower AI-enabled delivery costs, before the market fully adjusts.
  • Buyers already spending money on these services.
  • Growing competitive pressure on businesses that have not adopted AI.
• Operational principle: Build an AI-native, agile team, but keep trained people involved where human judgment and quality assurance matter.

Metrics and examples cited

Measure Earlier agency model AI-enabled model described Monthly price Typically $2,500–$3,000; broader range $1,500–$3,000 $300–$800 Customer acquisition cost (CAC) About $1,500 About $300–$400 Close rate About 30% Described as higher, with no specific figure Delivery time per client More than 10 hours/month Less than 1 hour/month Gross margin Roughly 50–60% Roughly 95–99%

• Dunn says traditional local agencies often lose clients after three to four months. At a $3,000 monthly retainer, that would mean roughly $9,000–$12,000 in revenue per client over that period.
• He gives a marketing example in which AI reduces the cost of a service previously priced around $3,000 to a possible $800 offer. He also says his own service company has moved retainers from $3,000 to $800, and in some cases $300.
• In a real-estate agency he sold in 2020, the work included ads, media buying, landing pages, and a call center. Dunn says client acquisition and fulfillment costs could make the first month unprofitable, with later months producing profit if the client stayed.
• For local home-service businesses, he suggests buyers may be less able to produce comparable marketing work themselves. This may allow providers to charge around $1,000/month for content and social media services, even when AI delivery costs are very low.
• He cites a market estimate of $4.6 trillion for services-as-software opportunities and says companies spend about $6 on services for every $1 spent on software. These figures are presented in the video as supporting context.

Risks and execution considerations

• Commoditization and price pressure: Visible deliverables can be copied or undercut, especially as more providers adopt AI.
• Customer self-service: Some buyers may try to do the work themselves. Dunn says he has encountered this objection, but claims clients often recognize a quality gap when comparing their work with a specialist’s.
• Quality and retention: Automating without supervision can reduce quality and increase cancellations. Human oversight remains important.
• Human bottleneck: Dunn estimates that a small portion of delivery still needs people. His ideal service business is lean and AI-native, rather than fully automated.
• Acquisition requirements: Lower prices can expand the market but reduce revenue per customer, making efficient, repeatable acquisition essential.
• Limited timing: He estimates the current arbitrage could last roughly 24 months, and until around 2028 in markets such as the US, UK, and Canada. He expects adoption to take longer in some other countries and languages.

Sources and presenter

• Presenter: Andrew Dunn.
• Organizations and companies referenced: Foundation Capital; Y Combinator; Andreessen Horowitz (a16z); Sequoia; Harvey AI; and “LawCoia” (name as rendered in the subtitles).
• Other person referenced: Alex Hormozi, mentioned by Dunn as a personal credential.